Time invariant steady-state Kalman filter for image super-resolution

Abstract

The Kalman filter is usually considered to be too computationally infeasible for image super-resolution due to the size of the images and thus the state space involved. This paper presents a computationally feasible, simplified time invariant steady-state Kalman filter approach to image super-resolution for the case of constant translational camera motion between low-resolution frames. Quantitative and aesthetic comparisons are made to the maximum a-posteriori (MAP) super-resolution approach, considered to be one of the standard super-resolution techniques, to demonstrate the Kalman filter's superiority despite the steady-state assumption.